moving X11 portforwarding out into a & quot; plugin& quot; framework

Brian J. Murrell brian at interlinx.bc.ca
Mon Jan 25 01:20:06 EST 2010


Alex Bligh <alex <at> alex.org.uk> writes: 
> 
> That's what I thought. And there is no reason in principle why other
> tunneling technologies shouldn't also be vendor extensions, as opposed
> to standards

Yeah, it seems pretty heavy-weight to have to write a standard for every
protocol that might want to tunnel over SSH.  Now, what there might be a good 
case for 
standardizing is the facility that I am proposing, the one that allows for the 
specification of tunnels and configuration (on both ends) of, for example, 
environment variables, the creation of sockets perhaps, etc.

> (not that I quite understand what Brian is trying to do).

Well, there are a number of examples, off the top of my head, of protocols that 
could benefit from being tunneled, from a remote machine to a local machine, 
running a gnome desktop, for example.  Pulseaudio is one, dbus is another.
There are likely others, perhaps more application specific even.  But the idea 
that every application writer needs to go through a standardization process as 
well as hacking the openssh code directly just seems, IMHO, silly.  Rather,
there should be this framework that these application vendors, or O/S 
distributors perhaps, can utilize to get their protocols forwarded.



More information about the openssh-unix-dev mailing list